Mini-Experiments
Phone App Surveying
I conducted these mini-experiments to test out some of the thoery & find real world applications.
What datum does a professionals GNSS reciever use?
This experiment was done to illustrate how costly mistakes can be made by surveyors and engineers if they do not check settings on their instruments and consider what datums they should be using.
Professional GNSS equipment was used to survey permanent marks which had known coordinates and elevations. The measurement data from the receivers was compared to the reported coordinates from the Queensland Government.
The experiment proved that the WGS84 datum is close to the local datum GDA94 in the Gold Coast but also proved that care should be taken to check settings if highly accurate GDA94 results are required. It was also shown that GNSS receivers return ellipsoidal elevations which could cause a discrepancy of up to 70m in Australia if the user was unaware of the need to apply the separation values.
The receivers were an engineer’s Javad Triamph-2 and a surveyor’s Leica GPS1200. The receivers returned coordinates for horizontal position in terms of the WGS84 datum. These were within 0.04- 0.1 seconds of latitude and longitude when compared to the reported value which was on the local datum GDA94. (Note that the accuracy is effected by various other factors). Both recievers had a large range of options for choosing other datums.
The receivers both returned ellipsoidal heights which are approximately 40m different to the height above sea level on the local datum. The surveyor’s equipment had the ability to apply separation values to calculate height on datum (in this case AHD) in the receiver or during the data processing stage using the accompanying software on the office computer. The engineer’s instrument did not appear to offer an automatic conversion to AHD and the user had used Geoscience Australia in the past to obtain the corrected values.
Users need to be aware of the likely need to apply the correct separation value their height measurements could cause problems.
Questions:
1. The earth is moving with continental drift and earthquakes, and we have maps with lots of different datums but why don’t the general public notice the coordinates changing?
2. Will we notice changing coordinates on everyday maps?
To answer at these questions the location of Brisbane was estimated from some different maps.
Map 1: the oldest map from 1859 which appears to show Brisbane at approximately 27˚30’, 153˚10’. The scale of the map is 35miles to an inch, each grid is just over 50miles wide.
Map 2: a map of Australia from the Collins atlas (1983). Brisbane is shown at approximately 27˚24’, 153˚03’. The scale is 1:12,000,000, 1cm represents 120km when printed.
Map 3: a map of eastern Australia from the same atlas. Brisbane is shown at approximately 27˚25’, 153˚02’. The scale is 1:6,000,000, 1cm represents 60km when printed.
Comparison with some commonly used websites:
Wikipedia defines the latitude and longitude of Brisbane as 27˚28’S, 153˚02’E.
Google Earth shows 27˚28’S, 153˚02’E when the mouse is hovered over the city.
Results
Discussion of results:
As Australia moves north easterly approximately 7cm a year due to continental drift it would be expected that the coordinates would change from old to new maps in this direction.
However this investigation has proven that the scale of the map is the reason that small scale movements of land are not noticeable on our maps. Over 156 years from the date of the first map, the country could have moved 1092m. As one degree on latitude is approximately 111km so a movement of 1-2km would not be noticed at a large scale.
The shift is noticeable on smaller scale maps such as topographic hiking maps and navigational charts.
It was surprising to see that all of the maps gave similar coordinates for Brisbane especially due to considerations of different accuracies for cartographers based on their available tools.
The reason for this is that the coordinates are approximate to the nearest minute and the error range is 8minutes. One minute difference in latitude and longitude in this region is approximately 2.5km so up to 20km of potential error, which given the scale of the maps viewed and the lack of defined point on the ground (the map symbol for Brisbane city was used to measure too), is surprisingly good.
The scale of the map can also hide the difference from AGD to GDA as shown in the table above. Points on a map with a scale of 1:5,000 will likely be around 40 mm different.
Try it out - get an old street directory (some have coordinates in E&N) or other map which is made in a datum other than WGS84 and compare the position of features to that shown on Google Earth.
Comparing old & new maps. Did positions change?
Showing that 0 degrees Longitude is not always over the line in Greenwich
This test shows the difference between traditional mapping using astronomical reference systems and the modern GNSS systems by using the location of the central meridian in London.
The central meridian representing 0 degrees longitude at Greenwich, was located on Google Earth using imagery. Then the location of 0 Longitude found using the coordinate values on the Google Earth Screen. The distance as shown in the image below, measured between them was approximately 103.5 meters which is close to the reported difference of approximately 100 meters (National Maritime Museum 2015).
Photo showing the difference between Longitude 0 in relation to astronomical and WGS84 datums at Greenwich London (Google Inc. 2015).
Aims:
A. to find out the accuracy of locational android apps.
B. to see the difference between different datums.
C. to assess the accuracy of elevation results from apps.
Method:
1. Choose some apps for android which are a mix of basic-advanced use and are free to download. Some apps offer elevations, some position, some both.
2. Choose permanent survey marks (PSM) which are in open sky areas with known accuracy for position and elevation.
3. Set up the android device – Samsung Galaxy S6, by turning on the position feature and internet. (4G internet was used for the experiments). Leave the phone for a few minutes.
4. Open an app and set it over a PSM with the centre of the phone over the mark. Leave the phone for 2minutes to allow the app to improve accuracy.
5. With some of the more advanced apps, the readings were then done in multiple different datums.
6. Compare the position and elevation readings to the reported coordinates from the Queensland Government Department DNRM.
Some of the limitations to consider:
The phone is approximately 145mm by 75mm in size and the antenna point was not considered. This creates a possible position error of 145mm and a vertical error of 7mm which is the width of the phone.
The internal workings of the phone, like any GNSS capable device, limits the accuracy achievable to certain levels. The phone was not designed to produce survey quality measurements but general users would expected to be within 10meters.
General GNSS survey limitations apply such as multipath, signal block from trees or buildings and satellite geometry. The position chosen for this experiment had a clear sky above and clear space around 3 quadrants around it. There were some trees to the south east of the mark approximately 10meters away.
Results
The apps performed better than expected giving coordinates within 5meters of the official GDA94 position as reported by the Queensland Government. The best app was ‘Handy GPS’ which was within 0.7m. It was also interesting to see that the coordinates viewed on Google Earth with the Queensland Globe layer were within 0.136m.
The apps were not as good with elevation calculations. Those which showed an elevation above mean sea level were 5-12m wrong. ‘Accurate Altimeter' was the closest but it is suspected that it obtained the elevation value of 2m from Google Earth via internet connection. Google earth shows an approximate elevation at ground while the apps were showing height at the PSM which is on top of a pillar of approximate height 1.3m. Therefore the AHD height at ground is approximately 0.893m which makes the Google Earth height of 2m approximately 1.107m different to AHD.
The apps which showed ellipsoidal elevations were 2.5-9.2m different to the AHD value. ‘Handy GPS’ was again the best at determining elevation as when the advertised accuracy of +/-4m is taken into account, the elevation range covers the AHD value making the reading potentially correct.
The most important finding from this experiment was finding that the apps do not give users much support in understanding the measurement values delivered. ‘Handy GPS’ and ‘Mobile Topographic Mapper’ gave options to change datums between WGS84, AGD and GDA (Mobile Topographic did not have the current GDA datum available) but there was no information about what these datum settings mean. It is expected that users have prior knowledge. The other apps have no information about datums nor do they offer any option to change datum. It is assumed that they are based on WGS84 as the results indicate this.
With the introduction of GDA94 being closely aligned to WGS84, this is not a likely to be a problem for recreational users who require accuracy of up to 5m. A lack of information is also an issue for elevation/altitude readings which are not only inaccurate but more importantly often do not advertise if they are ellipsoidal elevations or mean sea level. None mention AHD or AUSGeoid as an option in settings. ‘Handy GPS’ does offer the option of manual entry of a geoid correction value.
Conclusion
Positioning apps for devices such as smart phones and tablets are good tools for positioning using GNSS satellites but some are better than others. Many achieved horizontal positioning, in this particular environment, of 0.7-2meters. The apps worked better than expected and are great for their intended non-professional use.
Elevation determination was not so accurate, with the best being approximately 3meters different to the ellipsoidal height. More importantly there was little information for the user about the type of elevation being displayed. The majority displayed ellipsoidal heights which at best is 40.6m higher than the AHD value in this area.
Overall ‘Handy GPS’ provided the best positioning results for both coordinates and elevation with options to change datums.
The difference between datums was proven as the coordinates measured using GDA94 or WGS84 were within 1-2 meters of the recorded location in the permanent mark report. 1-2 meters is acceptable accuracy given the equipment used.
The coordinates calculated using AGD66 and AGD83 were 208-213m different to the official coordinates (GDA94) which is consistent with the expected difference of approximately 206m (ICSM, 2015).
When choosing a positioning app look for one which explains the datums it is using and the accuracy to expect - or just test it out before relying on it.
References for Apps
AR labs (2015) Accurate altimeter free v 1.15. [Android app] http://www.arlabs-mobile.com/apps/altimeter.html
Dunk, A., (2015) Binary Earth, Handy GPS. V 11.6. [Android app] http://www.binaryearth.net/HandyGPS/index.php
StgrDev. (2015) Mobile topographer free v 7.3.0. [Android app]
GPS Tools. (2015) My GPS altitude v.1.4. [Android app]
Woozilli Inc. (2013) GPS Coordinates v.1.1. [Android app]
Watch out for Map Distortions
"Projection" causes maps to become distorted in different ways.
When looking at maps of the world consider what tricks the map is playing on you.
One example is the distortion shown here on a wall map of the world. Notice that the lines of Lattitude are wider in the north than south. This makes the countries in the far northern lattitudes look big compared to their actual size relative to those in the south.
The actual distance between lines of lattitude is equal, but on this map there is 3cm spacing over Australia and 11cm north of Alaska.
The map had to be drawn this way to create the round surface onto a flat surface using a 'Mercator Projection'
Here's the Central Meridian at astronomical 0 degrees longitude.
But down here on the coordinates it says 0˚00’04.9”
Map refereneces:
Gregorys UDB World Map 160, (2013) Mercator projection, equatorial scale 1:40,000,000
Collins International Atlas, (1980)